Hot Topics #21 (Mar. 17, 2023) 19 implied HN points • 17 Mar 23 🕹 Technology AI ML Optimization Speech Synthesis Robotics GPT-4 is a new large-scale model by OpenAI that can accept image and text inputs to produce text outputs. PaLM-E is an embodied multimodal language model that incorporates real-world sensor data into language tasks. Meta-black-box optimization can discover effective update rules for evolution strategies through meta-learning.
Hot Topics #18 (Feb. 14, 2023) 0 implied HN points • 14 Feb 23 🕹 Technology ML Bioinformatics Code generation Machine learning model predicts activation energies of hydrogen atom transfer in proteins CodeBERTScore evaluates code generation using pretrained models SWARM parallelism offers efficient communication for training large models
Hot Topics #19 (Feb. 21, 2023) 0 implied HN points • 22 Feb 23 🕹 Technology ML Robotics Models Optimization Algorithms Molecular optimization using multi-objective Bayesian optimization and GFlowNets. Discovery of a simple and effective optimization algorithm, Lion, for deep neural network training. DreamerV3 algorithm based on world models outperforms previous approaches in various domains.
Hot Topics #22 (Apr. 3, 2023) 0 implied HN points • 03 Apr 23 🕹 Technology ML Optimization Language Models Machine Learning Robotics Fast Imitation of Skills from Humans (FISH) can train robots with less than a minute of demonstrations. Regularization and Lipschitz regularization are key in Optimal Transport-Based Distributionally Robust Optimization. Chain of Hindsight technique helps align language models with human preferences by training on feedback sequences.
Hot Topics #20 (Mar. 1, 2023) 0 implied HN points • 01 Mar 23 🔬 Science Deep Learning Neural Networks Protein design using deep learning techniques to create custom biocatalysts Efficient de novo protein design through relaxed sequence space for better computational efficiency Improving robotic learning with corrective augmentation through NeRF for better manipulation policies
Hot Topics #17 (Feb. 6, 2023) 0 implied HN points • 06 Feb 23 Hierarchical reinforcement learning is proposed for producing human-interpretable and generalizable program policies. Large language models can generate functional protein sequences across diverse families. SingSong is a system that generates musical accompaniments from singing using advanced source separation and audio generation techniques.
Hot Topics #23 (May 2, 2023) 0 implied HN points • 02 May 23 🕹 Technology ML Boosted Prompt Ensembles can enhance large language models' performance for reasoning Large language models like ChatGPT can excel in relevance ranking for Information Retrieval tasks Autonomous driving systems can be trained efficiently using deep RL without simulation or expert demonstrations